Bin Ding

Bin Ding#

Bin Ding

Bin Ding is an AI Engineer in the AMD AI group, specializing in developing and optimizing HIP and triton kernels on AMD Instinct GPUs for both inference and training. In the past few years, he has a lot of experience in operator development and optimization. He has developed and optimized the forward and backward kernels of operators including Convolution, Flash Attention, and Native Sparse Attention. He is very interested in how to use various optimization methods to fully utilize the ultimate capabilities of GPU hardware.

Posts by Bin Ding